Frequently asked questions

Joghd reads your content — website, documents, files — chunks and embeds it, and builds a conversational agent that answers strictly from that material. You then publish the agent to your website, Telegram, Bale, or through the API.

For a typical website, under ten minutes: paste the URL, crawling and embedding run automatically, and the widget snippet is ready. Larger document sets take a few minutes longer.

The agent doesn't invent one. If no chunk clears the confidence threshold it says so plainly, and — if you've configured it — hands the conversation to a human operator.

The interface is Persian and English, and the agent replies in the language of the source document. Retrieval is specifically tuned for Persian: ye/ke normalisation, diacritic stripping and half-space handling.

No. Your content is used only for retrieval inside your own workspace and never enters any model's training set.

On infrastructure inside Iran. On-premise deployment is available for enterprise customers.

Yes. Deleting a knowledge base purges its vectors too, and deleting a workspace removes everything. Conversation export is available at any time.

Yes. 50 AI messages, one agent and one knowledge base, with no card required and no time limit.

Through a domestic payment gateway using Iranian bank cards. A formal invoice is issued for every paid plan.

At any time. Upgrades apply immediately and are pro-rated. There is no mandatory contract and no cancellation fee.

No. The widget script is under 15 KB, loads asynchronously, and pulls nothing heavy until a visitor actually opens it.

Yes — a full REST API plus webhooks, available on Professional and above, so you can embed the agent inside your own product.